eines möchte ich gleich vorausschicken: die infos hier sind seeehr technischer natur und daher vielleicht nicht für jedermann verständlich aber die grundmessage kommt weiter unten doch gut rüber
http://www.nvnews.net/vbulletin/sho...mp;pagenumber=2einige auszüge aus dem nvnews forum wo bekanntermaßen uttar und hellbinder ziemlich zuverlässige quellen für grafikkarten specs sind:
Uttar:
" The NV3x supports FP32, FP16 and FX12 ( the NV35/NV36/NV38 does not have dedicated FX12 hardware however )
- The NV4x supports FP32, FP16 and FX16. FX16 *can* be used at places where FP16 would be insufficent. Ain't gonna go into the details, but a good example is PS1.4.: Theorically PS1.4. on current nV cards should get FP32, while on the NV4x, it should get FX16.
- Full parallelism between FP and Texture ops seems unlikely to me. Some things can be done for reducing the problem there though IMO...
- If some things I recently heard are true, the NV4x might win the AA IQ round this time.
darauf Hellbinder:
"Ok now I have my doubts that Nvidia has a grip at all.
There is absolutly no reason for them to Continue down this multi percision path. Its a complete waste of resources and it Divides their Transistor budget in ways that will ultimately hurt performance for them. This tells me that they are still going down a path not in line with any of the other IHVs or DX APi.
PS 1.4 seems to work just fine on ATi's Pure FP24 system. Its hardly "insufficient". Neither would it be in a Pure FP32 enviornment. I can see now that the Nv40 is NOT a new design and that it is indeed nothing but a tweaked up Nv30 on a much larger scale.
If what you believe to be true is true then I honestly Fear for them as a Company when they face Loki."
......
"It has occured to me now having seen this from Uttar that this is also why they are not backing down on their Nv35 hacks at all. Why they are still playing the PR mudslinging, coverup, Special deals with developers Game.
The Nv40 is just another Extension of the Nv30 except on a larger scale. It's going to still need the same kind of "Optimizations" as the current line up.
I now wonder if everyone at Nvidia has gone Completely insane."
und bitte diese GERÜCHTE auch als solche ansehen
3Dfx
Anmeldung